With the rapid adoption of and innovation around artificial intelligence (AI), the entire healthcare industry is on the cusp of transformation with new clinical and administrative use cases for emerging technology. However, it is critical that as AI continues to evolve, new applications and their associated benefits are being translated to the clinic and deliver value directly to patients.
Lung intervention is one of the pulmonology areas where AI imaging will play an increasingly important role, having significant impacts on patient care. Already today, AI aids lung care specialists in the in-depth analysis of thoracic CT images, providing precisely quantified data about lung anatomy, enabling more accurate diagnosis and planning of local therapies. These include, for example, bronchoscopic lung volume reduction in COPD patients with severe emphysema or lung nodule biopsy.
At the same time, there are a multitude of studies pending worldwide, showing immense potential opportunities to advance pulmonary interventions by minimizing the invasiveness while maximizing patient outcomes and safety, in a variety of lung therapies.
Looking inside the lungs
Lungs are one of the most complex human organs and structural changes caused by various diseases can be challenging to detect. Imaging using computed tomography scans (CT) is an important tool for clinicians to visually examine the anatomy of the lungs. However, the analysis of CT images still has important limitations.
There is a wide range of different structural changes in the lungs that can be caused by the underlying disease. On the other hand, many lung diseases are characterized by several similar abnormalities in airways, parenchyma or pulmonary vasculature. In addition, there are also multiple anatomical variations between patients, adding to the complexity of interpreting lung CT scans. Furthermore, the patient’s level of inspiration, and the technical aspects of CT acquisition like radiation dose and selected reconstruction kernel, can all affect the visibility of the smaller bronchi, and the visualization of the lung parenchyma on the scan. For this reason, visual examination of these CT scans can be challenging, and relevant changes in the lung structures can be easily missed.
Given that clinicians can only assess what they can see on the scan, the assessment quality strongly depends on their experience. In any case, it remains mostly a subjective analysis, sometimes based on manual and time-consuming quantification methods. Image analysis techniques based on AI make it now possible to evaluate clinically relevant lung structures with great precision and accuracy.
“Humans are very good at making qualitative judgment, but the quantification part of it is very difficult. I see the synergy between healthcare professionals and AI technology, not as one replacing the other – but empowering the other to do better in their work” says Dr. Jean-Paul Charbonnier, Chief Innovation Officer at Thirona.
For interventional procedures, imaging is already an important component of pre-operative assessment and planning. In the case of bronchoscopic procedures, for example, such as lung cancer biopsy or ablation, imaging provides crucial information on how to access small nodules in peripheral lung areas and difficult to reach locations. Still, changes in representation of lung anatomy between the pre-procedural CT scan and the bronchoscopic procedure often though lead to a divergence between the expected and actual location of the target lesion during the intervention. Alongside the CT-to-body divergence can present further complications for clinicians navigating the lungs and to hit the target precisely. In fact, facing challenges in reaching specific locations, clinicians must often reexamine the scan during the intervention – which results in lung treatment relying on a certain degree of expertise-based guesswork. Applying artificial intelligence technology to analyze intra-operative imaging, in real time during the procedure, is another very promising development taking place in lung imaging.
How AI Is Transforming Today’s Interventions Landscape
By integrating analytical capabilities of AI with medical imaging, technology is revolutionizing the way pulmonologists can evaluate and treat patients. Quantifying lung anatomical structures from a CT scan more accurately and detecting details the human eye cannot see; AI aids doctors in the decision-making process. Furthermore, with the rise in the number of lung screening programs, an increased workload for radiologists and pulmonary interventionists across the field is to be expected. With the support of AI, doctors will be able acquire sensitive and accurate, clinically relevant data to provide patients with accurate diagnosis and the best possible treatment.
“In the course of the last couple of years Artificial Intelligence-enabled analysis of lung CT scans has already become a world standard for qualifying COPD patients with severe emphysema, for an endobronchial valve placement”, says Dr. Dirk-Jan Slebos, MD, Head of Pulmonary Medicine and Tuberculosis and a Professor of Pulmonology at University Medical Center Groningen in the Netherlands, specialized in interventional bronchoscopy. “Having anatomical lung structures quantified down to a few millimetres, and the vascular density precisely calculated by AI, I can confidently take the most optimal treatment decisions for my patient. This is the type of approach that will help us make the most out of AI, because it allows us to devote our human effort in the places it’s needed the most.”
Enabling a more accurate analysis of lung anatomy and precisely indicated navigational paths to diseased areas, AI carries an immense promise to transform patient treatment by offering more localized and tailored, therapies, including bronchoscopy and surgical interventions. With more precise delineation of the lung segments, doctors will be able to treat the specific lung segment in which a tumor resides instead of the entire lung lobe. Such an approach could result in less invasive and more effective procedures, saving more healthy lung tissue and minimizing harm to patients.
While AI is contributing to great strides in the field of pulmonology, the involvement of humans will always be essential in healthcare, but moving forward, so will the role of AI.
The Future Promise of AI in Driving Patient Value
AI technology is expected to accelerate innovation in many pulmonary interventions requiring high precision navigation on a subsegmental and segmental level. These include lung cancer biopsies, surgical lung volume reduction, lung segmentectomy, ablation procedures, and more. Looking forward, experts in the field anticipate significant opportunities to build on the value of AI-based image analysis in lung care.
Despite the proliferation of more advanced lung screening tools today that enable earlier detection of lung cancer, many clinical challenges still exist in pulmonary interventional medicine on its own. As multiple studies are pending worldwide, it’s only a question of time that the integration of AI will revolutionize the landscape of lung cancer treatment and its capacity to aid in real-time decision-making during the therapeutic procedures, ultimately streamlining the clinical pathway processes, improving patient outcomes, and equalizing access to optimal healthcare.
The integration of AI into lung imaging and image analysis is transforming the field, providing a deeper understanding of the personal variance in lung anatomy and offering unprecedented advancements in diagnostic accuracy, treatment efficiency, and patient outcomes. According to Prof. Dr. Pallav Shah, MD, interventional pulmonologist at Royal Brompton Hospital in London and a world leading bronchoscopist: “With the shift towards minimally invasive techniques for diagnosis and treatment of both malignant and non-malignant lung and airway disorders, I expect Artificial Intelligence to become an integral component of the real-time navigation guidance during the peripheral bronchoscopy procedures.”
As AI continues to evolve, becoming an indispensable tool in modern pulmonology, it is paving the way for more personalized healthcare in pulmonary precision medicine. Collaboration between AI experts and medical professionals to generate clinical evidence on improved outcomes and patient safety measures from the regulatory perspective, will remain critical in harnessing the full potential of AI, to ensure the benefits of the technology reach patients worldwide.
Eva van Rikxoort
Eva van Rikxoort is the founder and CEO of Thirona, an AI company that uses advanced technology combined with medical science to bridge the gap between pulmonary research and clinical care. Eva has her Master’s in Artificial Intelligence (Radboud University Nijmegen) and obtained her Ph.D. on Segmentation of Anatomical Structures in Chest CT Scans from the University Medical Center Utrecht, followed with a two-year postdoc position in quantitative image analysis at the University of California-Los Angeles.
In 2014, Eva left academia to found Thirona, and has partnered with drug developers, clinicians, and medical device companies to help accelerate research and make personalized treatment in lung diseases accessible for everyone.
